Post-Enumeration Survey Overview

Presentation to the National Advisory Committee on Racial, Ethnic and Other Populations

May 6, 2021

Tim Kennel, Assistant Division Chief, Statistical Methods, Decennial Statistical Studies Division

2020CENSUS.GOV Post-Enumeration Survey Outline

1.Introduction to Census Quality 2.2020 Post-Enumeration Survey Design 3.Schedule and Reports

2 2020CENSUS.GOV Post-Enumeration Survey

Introduction to Census Quality

3 2020CENSUS.GOV Post-Enumeration Survey Measuring Census Quality

• Census process indicators • Assessments and evaluations of census operations • Demographic Analysis (DA) and other methods • Post-Enumeration Survey (PES)

4 2020CENSUS.GOV Post-Enumeration Survey Coverage Measurement in the Past

Demographic analysis for decades Coverage measurement surveys • 1980 Post-Enumeration Program • 1990 Post-Enumeration Survey • 2000 Accuracy and Coverage Evaluation • 2010 Census Coverage Measurement survey

5 2020CENSUS.GOV Post-Enumeration Survey 2010 Census and Demographic Analysis (in Thousands)

2010 Census 308,746

6 2020CENSUS.GOV Post-Enumeration Survey Net Coverage Error

Net census coverage error (defined as) = Census count – True population size

• Negative coverage error means fewer people are included in the census than are in the true population (undercount) • Positive coverage error means more people are included in the census than are in the true population (overcount)

Note: In the past, that is, up through 2010, documentation of results from the Census Bureau’s post-enumeration surveys has generally defined the net undercount using the opposite of the difference above. When providing results from 2020, and in these slides covering past results, we will use the definition above.

7 2020CENSUS.GOV Post-Enumeration Survey Net Undercount of US Household Population

Year Census Count Net Coverage Error Percent Coverage Error

2010 300,700,000 36,000* 0.01 %* 2000 273,600,000 1,332,000* 0.49 %* 1990 248,700,000 – 3,994,000* – 1.61 %* * Statistically different from zero Census counts from 1990 to 2010 ,exclude Remote Alaska. Census counts for 2000 and 2010 exclude Group Quarters population. Census count for 1990 included people in housing units and in noninstitutional and nonmilitary Group Quarters.

8 2020CENSUS.GOV Post-Enumeration Survey U.S. Total Percent Net Coverage Error from 1990 - 2010

Percent Net Coverage Error 1.00%

0.50%

0.00%

-0.50%

-1.00%

-1.50%

-2.00% Error bars indicate 90% -2.50% confidence interval. 1990 2000 2010

9 2020CENSUS.GOV Post-Enumeration Survey Components of Census Coverage for the 2010 United States Household Population (in Thousands) Standard Standard Component of Census Coverage Estimate Percent Error Error Census Count 300,700 0 100.0 Correct enumerations 284,700 199 94.7 0.1 Erroneous enumerations 10,040 199 3.3 0.1 Due to duplication 8,521 194 2.8 0.1 For other reasons 1,520 45 0.5 0.0 Whole-Person Census Imputations 5,993 0 2.0 0.0

Estimate of Population 300,700 429 100.0 Correct enumerations 284,700 199 94.7 0.1 Omissions 16,000 440 5.3 0.1

Net Coverage Error 36 429 0.01 0.1

10 2020CENSUS.GOV Post-Enumeration Survey

2020 Post-Enumeration Survey Design

11 2020CENSUS.GOV Post-Enumeration Survey What is the Post-Enumeration Survey?

Goal How is coverage estimated? • Measure coverage of people and housing • Conduct a survey in sample blocks units in the decennial census • Match people in the PES independent survey with • Measure net coverage error and enumerations in the census to determine who was components of coverage missed or counted in error • No intent to adjust the census counts History How are the data used? • Used in the United States census since 1950 • To assess the success of the current census • To provide information to improve the coverage of future censuses

12 2020CENSUS.GOV Post-Enumeration Survey 2020 Post-Enumeration Survey Universe

Household Population and Housing Units Excludes • Group Quarters and Group Quarters Population • Remote Alaska

13 2020CENSUS.GOV Post-Enumeration Survey Sample Sizes

First sample (Basic Collection Units) • Selected 9,800 Basic Collection Units in 50 states and District of Columbia • Selected 400 Basic Collection Units in Puerto Rico PES listing • 510,000 housing units in 50 states and District of Columbia • 18,000 housing units in Puerto Rico After subsampling and review • 161,000 housing units in 50 states and District of Columbia • 7,100 housing units in Puerto Rico

14 2020CENSUS.GOV Post-Enumeration Survey Two enumerations in sample areas

Capture (Enumeration sample, E sample) • A sample of census records (enumerations) in the PES sample blocks, called E sample • Measure correct enumerations in census • Determine if enumeration is correct or erroneous; duplicate? correct location?

Recapture (PES sample, population sample, P sample) • A sample of blocks, independent of the census, in the post-enumeration survey (PES), called P Sample; interview people in units • Match people in P Sample to Census

15 2020CENSUS.GOV Post-Enumeration Survey How PES Estimates are Created

16 2020CENSUS.GOV Post-Enumeration Survey Current version of Dual System Estimator (DSE)

, = , , 𝜋𝜋ce 𝑗𝑗 �𝐷𝐷 𝑑𝑑𝑑𝑑 𝑗𝑗 m 𝑗𝑗 𝑇𝑇𝑇𝑇𝑇𝑇Where � 𝜋𝜋 �𝜋𝜋 𝑗𝑗∈𝐷𝐷 j is a census enumeration, and D is the estimation domain (total, tenure status, age/sex groupings, etc.).

πdd,j is the predicted probability census case j is data-defined

πcd,j is the predicted probability that census case j is a correct enumeration

π m,j is the predicted probability that census case j was matched

17 2020CENSUS.GOV Post-Enumeration Survey

Schedule and Reports

18 2020CENSUS.GOV Post-Enumeration Survey Revised field milestone schedule

Original Dates New Dates Operation Start Finish Start Finish Independent Listing 1/16/2020 3/20/2020 1/16/2020 3/20/2020 Initial HU Followup 5/6/2020 6/19/2020 7/30/2020 9/21/2020 Person Interview 6/17/2020 9/18/2020 9/23/2020 3/20/2021 Person Followup 2/3/2021 3/26/2021 6/14/2021 8/12/2021 Final HU Followup 5/19/2021 6/18/2021 10/25/2021 11/30/2021

Operations in blue are complete. Dates include reinterview.

19 2020CENSUS.GOV Post-Enumeration Survey Tentative Person Tabulations from PES

Three major groups: • Geographic areas • Demographic or housing unit characteristics • Region • State • Age Group • Census operations • Age Group by Sex • Coverage Improvement Universe • Sex • GSS Partnership County • Race and Hispanic Origin (alone or in- • In-Field Address Canvassing Outcome combination and mutually exclusive • Nonresponse Followup Respondent Type groups) • Nonresponse Followup Workload • Relationship to Householder • Response Mode • Tenure (owner or renter) • Response Rate Decile • Type of Enumeration Area

20 2020CENSUS.GOV Post-Enumeration Survey Release Coverage Reports and Tables

First Release • National demographic estimates of net coverage of people • National demographic estimates of components of coverage for people • Tables can be downloaded on Census Bureau website

Second Release • State coverage estimates for people • More detailed national coverage estimates for people • All housing unit coverage estimates • Coverage estimates for Puerto Rico

21 2020CENSUS.GOV Post-Enumeration Survey Schedule to Release Methodology Reports

First Release • Source and Accuracy Statement – National Person

Second Release • Estimation Design • Estimation Methods: Characteristic Imputation • Estimation Methods: Missing Data • Estimation Methods: Net Coverage Estimation • Estimation Methods: Component Estimation • Source and Accuracy – All of PES

22 2020CENSUS.GOV Post-Enumeration Survey References • Mule, T. (2012), “2010 Census Coverage Measurement Estimation Report: Summary of Estimates of Coverage for Persons in the United States,” DSSD 2010 Census Coverage Measurement Memorandum Series #2010-G-01. • The National Academies of Sciences, Engineering, and Medicine. (2015). The Bicentennial Census: New Directions for Methodology in 1990. 30th Anniversary Edition. C.F. Citro and M.L. Cohen, Eds. Committee on National Statistics, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press. • U.S. Bureau of the Census. 1960 Censuses of Population and Housing: Procedural History. U.S. Government Printing Office, Washington, D.C., 1966. • U.S. Bureau of the Census, U.S. Census of Population and Housing: 1970, Procedural History, PHC (R)-1, Washington, D.C., 1976 • U.S. Census Bureau, 1980 Census of Population and Housing, History, Part E, PHC80-R-2E (Washington, DC: U.S. Government Printing Office, August 1989), • U.S. Census Bureau, 1990 Census of Population and Housing, History, Part D, 1990 CPH-R-2D (Washington, DC: U.S. Government Printing Office, March 1996). • U.S. Census Bureau, History: 2000 Census of Population and Housing, PHC-R-V2 (Washington, DC: U.S. Government Printing Office, December 2009).

23 2020CENSUS.GOV Post-Enumeration Survey Questions for NAC

How have you used census post-enumeration survey estimates in the past?

How do you think the Census Bureau should use the estimates of net coverage error and components of coverage in planning for the future?

24 2020CENSUS.GOV Questions

2020CENSUS.GOV